Fault Diagnosis of Knee Joint Based on Adaptive NEURO-Fuzzy Inference System

Vibroarthrography (VAG) is a simple and non-invasive technique for fault detection of knee joint. This paper presents the application of adaptive neuro-fuzzy inference system (ANFIS) model for classification of VAG signals. The approach consists of two stages. First, the diagnostic features are extracted from these signals using the wavelet transform (WT). Secondly, an ANFIS based expert system is developed for classification. A zero-order Takagi-Sugeno model is chosen for ANFIS architecture. The back-propagation gradient descent method in combination with the least squares method is adopted for training of expert system. The system is trained using a set of 16 normal and abnormal signals. The performance of the ANFIS model is evaluated with different 46 VAG signals in terms of sensitivity and overall accuracy. Compared with the other methods, ANFIS is able to achieve overall accuracy of 93.2%. The result confirmed that the proposed ANFIS model has potential to identify knee joint faults.